CS 481 - Team 49
Meeting Minutes

Weekly Team Meeting
Friday, October 19, 2007
10:30 - 11:30 AM
JEB 324
Recorder: Brandon Riggers
--------------------------------

Meeting Summary
   This meeting invovled discussion with Bruce about the 
command line options of our tool, the file arguments our 
tool will take, and our design.  We also talked about the 
separation of "run_turmoil" and "TurmoilBetweenLabels", and 
a "Unix Services for Windows" type of tool - which is similar 
to Cygwin in concept, and something we'll need to look into.  
Bruce concluded the meeting by mentioning scaling with their 
historical data.



Attendees
   Brandon Riggers, Peter Brown-Hayes, Travis Thoroman, 
Bill Junk, Bruce Mayes (over the phone), Shruti Upadhyaya 
(for brief portion)



Topics Discussed
   The meeting began with Bill getting our achievements 
for the week before Bruce called.  He also commented 
briefly on our Operational Specification, saying that we 
should just give an example string of the command line, 
instead of using a wordy description.
   When the teleconference began, Travis opened a 
discussion of the command line, asking about a "-p 
<product>" option Bruce expressed in an email.  Bruce 
said that we wouldn't worry about this option right now, 
but he can envision it as part of the tool after our work 
on the project is over.  Our tool should do nothing 
with "-p" for now.
   Travis also brought up the possibility of ambiguity 
with the user specifying simply a file and its revision 
number on the command line, and started discussion of 
determining the difference between a branch and a trunk in 
the repository.  Bruce talked about how we could pick 
between the following two ways of specifying files 
(these are examples from an email between Travis and Bruce): 

$ run_turmoil -o 9 -r 23 -l filelist https://Server.com/Source/

$ run_turmoil -o https://Server.com/Source/trunk:9
 -r https://Server.com/Source/branch:23 -l filelist

Bruce said we could choose the one that's easiest to 
implement, and Travis added that we'll have to pass the 
whole path to Subversion anyway.  Bruce also brought up 
using "svn cat" as an alternative to "svn checkout" when 
our tool gets the necessary files from Subversion.  Peter 
concluded the topic of the operation of the tool by 
confirming that we can assume the files given are just 
going to be Subversion files (i.e. we won't have to worry 
about someone passing the tool files not associated with 
Subversion).
   We talked briefly about the design of the tool, and the 
diagram we produced during the previous week.  Bill 
explained that our diagram separates the functionalities 
our tool performs, but "shouldn't imply packaging," or the 
way the tool is actually split up.
   Bruce then brought up the difference between the
"run_turmoil" and "TurmoilBetweenRevisions" portions of the 
tool.  The format of the output is really the only 
difference, and Bruce said he "is comfortable with having 
just one tool," with the different response of the tool 
being "called as A vs. if it's called as B."
   We then talked about the "Unix Services for Windows" 
tool, which is installed on HP's machines.  Bruce said 
that their builds take advantage of this tool and of using 
the 'diff' utility it provides.  We should run our tests 
using this version of 'diff.'
   Bruce asked about our level of work on the 
scalability factor with Linux diff and HP's historical data. 
Bill talked about how figuring out the scaling may be more 
on Bruce than us, as it's "potentially" his responsibility, 
and Bruce will have to run our tool on his stuff to "decide 
for himself" anyways.
   Lastly, Bruce confirmed our release schedule, asked if 
we'll be able to hit these deadlines - including all the 
deliverables with a release, such as documentation and 
tests - and confirmed our notion of scope with testing. 
One of the last things Bruce said as the meeting closed, 
was that "overall, this feels good."